Model comparison
canary-1b-flash vs cohere-transcribe-03-2026
At Q4_K_M, canary-1b-flash is the smaller download — 677,141,280 bytes against 1,558,162,944.
From the file· summed bytes, KV per layer
Side by side
| canary-1b-flash | cohere-transcribe-03-2026 | |
|---|---|---|
| Parameters | 811M | 2.1B |
| Architecture | canary | cohere_asr |
| Layers | — | — |
| Native context | — | — |
| Mixture of experts | no | no |
| Quantizations published | 6 | 11 |
| Smallest quantization | 0.63 GiB | 1.41 GiB |
| Q4_K_M | 0.63 GiB | 1.45 GiB |
| Licence | cc-by-4.0 | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | canary-1b-flash | cohere-transcribe-03-2026 | Ratio |
|---|---|---|---|
| 4,096 | — | — | — |
| 8,192 | — | — | — |
| 16,384 | — | — | — |
| 32,768 | — | — | — |
| 65,536 | — | — | — |
| 131,072 | — | — | — |